Discovery and structure-activity relationship of coumarin derivatives as TNF-α inhibitors

Article abstract of DOI:10.1016/j.bmcl.2004.03.022

The discovery and structure-activity relationship of a novel series of coumarin-based TNF-α inhibitors is described. Starting from the initial lead 1a, various derivatives were prepared surrounding the coumarin core structure to optimize the in vitro inhibitory activity of TNF-α production by human peripheral blood mononuclear cells (hPBMC), stimulated by bacterial lipopolysaccharide (LPS). Selected compounds also demonstrated in vivo inhibition of TNF-α production in rats.

Full text of DOI:10.1016/j.bmcl.2004.03.022

Bioorganic & Medicinal Chemistry Letters 14 (2004) 2411–2415
Discovery and structure–activity relationship of coumarin
derivatives as TNF-a inhibitors
Jie-Fei Cheng,a,* Mi Chen,a David Wallace,a Sovouthy Tith,a Thomas Arrhenius,a
Hirotaka Kashiwagi,b Yoshiyuki Ono,a,b Akira Ishikawa,a,b Haruhiko Sato,b
Toshiro Kozono,b Hediki Satob and Alex M. Nadzana
aDepartment of Chemistry, Chugai Pharma USA LLC, 6275 Nancy Ridge Drive, San Diego, CA 92121, USA
bFuji-Gotemba Research Laboratories, Chugai Pharmaceutical Co., Ltd, 1-135 Komakado, Shizuoka 412, Japan
Received 5 February 2004; revised 3 March 2004; accepted 8 March 2004
Abstract—The discovery and structure–activity relationship of a novel series of coumarin-based TNF-a inhibitors is described.
Starting from the initial lead 1a, various derivatives were prepared surrounding the coumarin core structure to optimize the in vitro
inhibitory activity of TNF-a production by human peripheral blood mononuclear cells (hPBMC), stimulated by bacterial lipo-
polysaccharide (LPS). Selected compounds also demonstrated in vivo inhibition of TNF-a production in rats.
ꢀ 2004 Elsevier Ltd. All rights reserved.
Tumor necrosis factor a (TNF-a) is a proinflammatory
cytokine secreted by a variety of cells, including mono-
cytes and macrophages, in response to many inflam-
matory stimuli or external cellular stress.1 It is a key
cytokine in the inflammation cascade, causing the pro-
duction and/or release of other cytokines and agents.
TNF-a exerts its biological effects through interaction
with one of two ubiquitously expressed cell surface
receptors, TNFR1 (p55) and TNFR2 (p75). Binding of
TNF-a to its receptors initiates a series of parallel pro-
tein serine/threonine kinase cascades, which lead to the
activation of members of the MAP kinase superfamily.
It also causes activation of the transcription factors
NFjB and Jun Kinase.2 NFjB in turn regulates the
production of many proinflammatory cytokines includ-
ing TNF-a and related proteins that are elevated in
immunoinflammatory diseases.3 TNF-a levels and
NFjB transcriptional activity are controlled by a
reciprocal feedback loop. Since excessive or unregulated
TNF-a production has been implicated in mediating or
exacerbating a number of disease states, for example,
cachexia and sepsis,4 decreasing TNF-a levels or inhib-
iting NFjB transcriptional activation constitute valu-
able therapeutic strategies for the treatment of many
inflammatory, infectious, immunological, or malignant
diseases such as rheumatoid arthritis, psoriasis, and
inflammatory bowel disease.5
Three macromolecular TNF-a inhibitors, such as
Enbrelꢁ, a soluble TNF-a receptor (TNFR2) coupled to
Fc portion of IgG,6a mouse human chimeric anti-human
TNF-a antibody6b;c;d Remicadeꢁ and human anti-
human TNF-a antibody Adalimumab7 have been shown
to be useful for the treatment of inflammatory and
autoimmune diseases. They are approved for reducing
the signs and symptoms of moderate to severe rheu-
matoid arthritis, psoriatic arthritis, and Crohn’s disease.
The efficacy of a number of small molecules7;8 with anti-
inflammatory activity has also been linked to their
ability to lower TNF-a levels.
As part of our continuing search for novel and potent
TNF-a inhibitors,9 we recently identified a coumarin-
based compound, dimethyl-carbamic acid 3-benzyl-4-
methyl-2-oxo-2H-chromen-7-yl ester (1a), as a moderate
inhibitor of TNF-a. Compound 1a inhibits NFjB and
AP-1 in a transfected cell based reporter gene assay as
well as TNF-a production in LPS-challenged human
peripheral blood mononuclear cells (1a: IC50 1.8 lM).
The aim of the present study was to establish the
structure–activity relationships for this class of inhibitor
and to identify derivatives with improved TNF-a
inhibitory activity.
Keywords: Coumarin; TNF-a.
* Corresponding author at present address: Tanabe Research Labora-
tories USA, Inc., 4540 Towne Centre Court, San Diego, CA 92121,
USA. Tel.: +1-858-622-7029; fax: +1-858-538-9383; e-mail: jcheng@
trlusa.com
0960-894X/$ - see front matter ꢀ 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.03.022

2412
J.-F. Cheng et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2411–2415
Synthesis of derivatives of compound 1 is straightfor-
ward. As shown in Scheme 1, the key intermediates, C-7
hydroxy coumarins 4, were prepared from substituted or
unsubstituted resorcinols and b-ketoesters according to
literature procedures.10 Thus, 2-arylmethyl b-ketoesters
3 reacted with resorcinols 2 in 70% sulfuric acid to
provide 3-arylmethyl-4-substituted coumarins 4. Upon
treatment with N,N-dimethylcarbamoyl chlorides in the
presence of NaH, the C-7 hydroxylcoumarins 4 were
readily converted into the desired N,N-dimethylcarba-
mates (1) in good to excellent yields (Scheme 1). Using
this procedure, a series of compounds with different
substituents, in particular halogen, cyano, nitro, lower
alkyl, trifluoromethyl, lower alkoxyl groups at C-5, C-6,
and C-8 positions, and alkyl/substituted alkyl groups at
C-3 and C-4 positions were prepared using a variety of
b-ketoesters and substituted resorcinols.
Coumarin derivatives shown above were evaluated in
LPS-challenged human peripheral blood mononuclear
cells (PBMC) for their ability to inhibit TNF-a pro-
duction, using an ELISA assay. The cells were cultured
in RPMI1640 medium supplemented with 5% heat-
inactivated fetal calf serum and antibiotics. PBMC
(5 · 105 cells/mL) in 0.2 mL aliquots were pretreated with
the test compound in DMSO for 60 min at 37 ꢂC in 96
well round-bottomed tissue culture plates. Thereafter,
the PBMC in the presence or absence of compound were
stimulated with 1 lg/mL lipopolysaccharide (LPS) from
Salmonella minnesota R595 at a final concentration of
100 ng/mL. After overnight culture, the supernatants
were harvested and assayed immediately for TNF-a.
The concentration of TNF-a in the supernatant is
determined using a human TNF-a ELISA Kit. A sec-
ondary assay using a reporter gene system confirmed
that these compounds inhibit NFjB and AP-1 activa-
tion (data not shown).
Alternatively, C-4 modified analogs could be synthesized
by elaboration of 4-methyl-courmain intermediate 4a
(R4 ¼ Me) as shown in Scheme 2. Deprotonation of
4-methyl-7-hydroxylcoumarin 4a with LiHMDS, followed
by treatment of the resultant anion with an aldehyde, an
alkyl halide, or an ester gave rise to the corresponding
hydroxyl, alkyl, or ketone products of the general
structure 5. Compound 5 was subsequently converted
into its corresponding N,N-dimethylcarbamate 1 under
the same reaction condition as shown in Scheme 1.
Modifications of the core coumarin structure of initial
hit 1a, such as removal of the 2-carbonyl group or 3,4-
double bond and replacement of 1-oxygen atom with an
NH group, did not generate active analogs. The N,N-
dimethylcarbamate at C-7 position could be replaced by
N,N-dimethylthiocarbamate or N,N-diethylcarbamate.
Other carbamates abolished the activity completely
(data not shown). Therefore the SAR studies were
focused on the substitutions around the coumarin core
with N,N-dimethylcarbamate at C-7 position. The
influence of substituents at the C-3 to C-8 positions of
coumarin was investigated. As shown in Table 1, a
hydrophobic group at C-3 appears to be preferred. For
example, compounds with a proton or a methyl group
at C-3 position did not show any TNF-a inhibitory
Scheme 3 describes the synthesis of C-3 benzyl substi-
tuted analogs. 3-Nitrobenzylcoumarin derivatives 7
were converted into their corresponding anilines by
hydrogenation. Reaction of the resultant aniline deriv-
atives 8 with an acyl chloride or an isocyanate provided
the final products 9.
R7
N
R8
R8
R8
O
O
O
O
O
HO
R6
O
O
HO
R6
OH
R7'
a
b
+
R4
OEt
O
R6
R3
R3
R3
R5 R4
1
R5 R4
4
R5
2
3
Scheme 1. Reagents and conditions: (a) 70% H2SO4, 100 ꢂC; (b) NaH, THF, R7R70NCOCl, rt.
HO
O
O
HO
O
O
X
b
a
N
O
O
6
O
X
Ar
Ar
O
Ar
R10
R10
4a
5
Scheme 2. Reagents and conditions: (a) LiHMDS, R10CHO, THF; (b) NaH, THF, Me2NCOCl.
HO
O
O
N
O
O
O
HO
O
O
b,c
NH2
a
O
R
R
NO2
NHCOR9 or
NHCONHR9
R
8
7
9
Scheme 3. Reagents and conditions: (a) H2/Pd–C, EtOH, rt; (b) R9COCl or R9NCO, THF; (c) Me2NCOCl, NaH, THF.

J.-F. Cheng et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2411–2415
2413
Table 1. In vitro inhibitory activity of C-3 modified coumarins on
TNF-a production
Table 2. In vitro inhibitory activity of C-4 modified coumarins on
TNF-a production
N
O
O
O
N
O
O
O
O
O
R3
R3
R4
Compounds
R3
Bn
IC50 (lM)
Com-
pounds
R3
R4
IC50
(lM)
1a
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
1.8
0.32
>50
P50
24.7
22.8
25.4
7.2
4-Fluorobenzyl
H
Me
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
Bn
Bn
Bn
H
Et
>50
1.7
0.78
n-Pr
n-Butyl
p-Fluorobenzyl n-Pr
1.2
2.1
n-Pentyl
n-Hexyl
c-Hexylmethyl
Bn
Bn
HOCH2CH2
COOH
>50
>50
1.7
p-Fluorobenzyl CF3
p-Fluorobenzyl Ph
1-Piperidinylmethyl
4-Morpholinylmethyl
Ph
>50
>50
>50
>50
6.5
Bn
Bn
Bn
Bn
Bn
Bn
Bn
Bn
Bn
Bn
Bn
o-F–Ph
p-F–Ph
4.9
41.7
Phenethyl
m-CF3–Ph
p-CF3–Ph
2-Pyridinyl
4-Pyridinyl
PhCH2CH2–
>50
>50
18.7
4.9
>50
2-Pyridinylmethyl
3-Pyridinylmethyl
4-Pyridinylmethyl
o-NH2Ph–
2.7
1.3
2.1
2.7
m-NH2Ph–
p-NH2Ph–
p-CNPhCH(OH)CH2–
2-Thiazolyl-CH(OH)CH2–
c-Hexyl-CH(OH)CH2–
28.3
16.5
5.2
7.5
o-MeOPh–
m-MeOPh–
>50
3.2
5.6
CH2@CH(CH2)2CH(OH)CH2– 48.4
p-MeOPh–
o-CF3Ph–
P50
>50
>50
1.7
m-CF3Ph–
p-CF3Ph–
methyl group (37) completely abolished the activity.
Compounds with C-4 ethyl (38), n-propyl (39, 40) or
hydroxyethyl (41) groups showed comparable activity to
the parent compound 1a. Replacement of C-4 methyl
group with a carboxylic acid (42) or a trifluoromethyl
(43) group resulted in inactive compounds. A phenyl
group is tolerated at this position (44: IC50 1.7 lM).
However, a substituent, such as a fluoro or trifluoro-
methyl group, at the para-position of the phenyl ring
dramatically decreased activity. Similarly, pyridinyl
groups (49, 50) were not good surrogates of a methyl or
ethyl group at this position.
m-MeO2CPh–
p-MeO2CPh–
m-AcNHPh–
m-EtNHCONHPh
P50
1.1
2.4
activity. The activity appears to increase with the size of
the alkyl group at this position, especially with carbo-
cycles (e.g., compound 16 with cyclohexylmethyl sub-
stitution). However, its activity is five times lower than
the aromatic counterpart 1a. Non-carbocyclic analogs
such as 4-morpholinylmethyl or 1-piperidinylmethyl
groups were not tolerated at the C-3 position.
Although a hydroxyethyl group (41) appeared to be
tolerated at the C-4 position, all attempts to introduce
other moieties at this position failed to generate more
potent TNF-a inhibitors. Those analogs with a phenyl
(51), substituted phenyl (52), heterocyclic (53), cyclo-
hexyl (54), or a linear aliphatic group (55) all showed a
decrease or complete loss of TNF-a inhibitory activity.
An aryl group greatly enhanced TNF-a inhibitory
activity, when attached to the C-3 position via a methyl-
ene link. Removal (19) or extension of this methylene
link to an ethylene (20) both abolished the activity. A
fivefold increase in TNF-a inhibitory activity was seen
when a fluoro atom was introduced to the para-position
of the C-3 benzyl group. However, a trifluoromethyl
group, a carboxylic acid, or an ester group at this position
was not tolerated. On the other hand, it seems that
modification of the meta-position of C-3 benzyl group
was well tolerated. Compounds bearing an acetamido
(35) or an ethyl ureido (36) group at the meta-position
were as potent as the parent compound 1a. Although the
4-pyridinylmethyl analog is the best TNF-a inhibitor
among the three possible regioisomers, all showed low
micromolar inhibition, similar to a benzyl group.
The most dramatic increase in TNF-a inhibitory activity
was observed for those compounds with C-6 substitu-
tions, especially with 6-halo substituted derivatives
(Table 3). For example, the 6-halo coumarin compounds
(63–65) prepared from 4-haloresorcinol was 20–30 times
more potent than the corresponding nonsubstituted
counterpart 1a. Alkyl groups except methyl group (56)
at this position tended to decrease the activities. No
clear trend was observed for electronic effects of the
substituents. Electron withdrawing group such as CHO
(60), CN (61), NO2 (62), and COOH (59) either
improved or diminished the TNF-a inhibitory activities.
Electron donating group like MeO (66) increased
A number of C-4 analogs (Table 2) were prepared and
tested. It appears that a small aliphatic group at the C-4
position is optimal for activity. Elimination of C-4

2414
J.-F. Cheng et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2411–2415
Table 3. In vitro inhibitory activity of C-5/C-8 modified coumarins on
TNF-a production
Table 4. In vivo inhibition of TNF-a production by coumarin com-
pounds
Compounds
Inhibition (%) @ 100 mg/kg sc
N
O
O
O
Control
10
––
42
86
57
86
99a
O
Ar
R
22
63
Com-
pounds
R
Ar
IC50 (lM)
67
Dexamethasone
a At 100 lg/kg.
1a
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
6-H
Ph–
Ph–
Ph–
Ph–
Ph–
Ph–
Ph–
Ph–
Ph–
Ph–
Ph–
Ph–
1.8
0.4
6-Me
6-Et
6-iPr
4.6
40.1
ꢀ50
9.8
6-COOH
6-CHO
6-CN
6-NO2
6-Cl
In summary, a novel series of coumarin-based carba-
mates was synthesized, which exhibited potent inhibi-
tory activity of TNF-a production in human PBMC
cells stimulated by bacterial lipopolysaccharide (LPS).
Substitution at C-6 position of the coumarin ring system
was found to most dramatically influence TNF-a
inhibitory activity. This observation led to the discovery
of 6-halo derivatives with low nanomolar activity. Sev-
eral of the coumarin derivatives (e.g., 22, 67) also
inhibited TNF-a production in rats. Several types of
coumarins have been reported in the literatures to
inhibit TNF-a production via different mechanism of
action.12 For example, cloricromene, a semi-synthetic
coumarin derivative, inhibited NFjB activation and
subsequent TNF-a neosynthesis, by scavenging reactive
oxygen species.12a;b More detailed mechanism of action
and pharmacological investigations of these compounds
may afford novel anti-TNF-a agents for the treatment of
autoimmunoinflammatory diseases.
0.85
0.12
0.09
0.06
0.06
0.58
0.45
0.13
ꢀ50
0.44
10.3
0.05
0.31
6-Br
6-I
6-OMe
6-Cl
6-Cl
3-Pyridinyl
4-Pyridinyl
Ph
5-OMe
5-F
Ph–
Ph–
8-Me
6-Cl
6-Cl
m-EtOCOCH2CONHPh–
m-AcNHPh–
activity by threefold, similar to the CN group. For
comparison, compounds with substitution at other
positions of the coumarin ring than C-6 were also pre-
pared. A C-8 methyl derivative (71) is 25 less potent than
the corresponding C-6 derivative (56). Similarly, a
methoxy group at C-5 position completely abolished the
TNF-a inhibitory activity. Only a small substituent like
fluorine (70) is tolerated at C-5 position.
References and notes
1. Locksley, R. M.; Killeen, N.; Lenardo, M. J. Cell 2001,
104, 487.
Improvement in potency was also observed for C-3
pyridinylmethyl or other substituted benzyl groups
when a halogen such as chlorine was present at the C-6
position. Pyridine derivatives 67 and 68 were about 6–10
times more potent than the corresponding nonsubsti-
tuted counterparts. Compound 73 (IC50 0.31 lM), which
possesses a C-6 chloro and substitutions at C-3 benzyl
group, is a better TNF-a inhibitor than 35 (IC50
1.1 lM).
2. Hsu, H.; Huang, J.; Shu, H. B.; Baichwal, V.; Goeddel,
D. V. Immunity 1996, 4, 387.
3. Palanki, M. S. S.; Erdman, P. E.; Gayo-Fung, L. M.;
Shevlin, G. I.; Sullivan, R. W.; Suto, M. J.; Goldman,
M. E.; Ransone, L. J.; Bennett, B. L.; Manning, A. M.
J. Med. Chem. 2000, 43, 3995.
4. Tumor Necrosis Factor; Beutler, B., Ed.; Raven: New
York, 1992.
5. Black, R. A.; Bird, T. A.; Mohler, K. M. Annu. Rep. Med.
Chem. 1997, 30, 241.
6. (a) Moreland, L. W. Exp. Opin. Invest. Drugs 1999, 8,
1443; (b) Mikuls, T. R.; Moreland, L. W. Exp. Opin.
Pharmacother. 2001, 2, 75; (c) Blam, M. E.; Stein, R. B.;
Lichtenstein, G. R. Am. J. Gastroenterol. 2001, 96, 1977;
(d) Assche, V. G.; Rutgeerts, P. Exp. Opin. Invest. Drugs
2000, 9, 103.
7. Palladino, M. A.; Bahjat, F. R.; Emmanuel, A.; Theodo-
rakis, E. A.; Moldawer, L. L. Nature Rev.: Drug Discov.
2003, 2, 736.
8. Thalimide and analogs: (a) Sampaio, E. P.; Sarno, E. N.;
Galilly, R.; Cohn, Z. A.; Kaplan, G. J. Exp. Med. 1991,
173, 699; (b) Niwayama, S.; Turk, B.; Liu, J. J. Med. Chem.
1996, 39, 3044; (c) Miyachi, H.; Azuma, A.; Hiioki, E.;
Iwasaki, S.; Kobayashi, Y.; Hashimoto, Y. Biochem.
Biophys. Res. Commun. 1996, 226, 439; (d) PDE-IV
inhibitors and MMP inhibitors: Nelson, F. C.; Zask, A.
Exp. Opin. Invest. Drugs 1999, 8, 383; (e) Lowe, C. Exp.
Representative compounds that possess excellent in
vitro TNF-a inhibitory activity were tested in vivo by
subcutaneous (sc) administration at 100 mg/kg (Table
4). TNF-a inhibitory activity was assessed by in vivo
inhibition of serum TNF-production in the rats.11 As a
result, compounds 10 and 63, which possess a C-3
benzyl group with IC50 at 0.32 and 0.09 lM, respec-
tively, showed only moderate inhibition of TNF-a
production in vivo. However, those derivatives with a 4-
pyridinylmethyl group at C-3 position (22: IC50 2.7 lM;
67: IC50 0.45 lM) demonstrated higher in vivo efficacy,
even though their in vitro activities are not as potent as
10 or 63. These results clearly indicated the importance
of solubility, along with the compound potency for the
in vivo TNF-a production inhibition.

J.-F. Cheng et al. / Bioorg. Med. Chem. Lett. 14 (2004) 2411–2415
2415
Opin. Ther. Pat. 1998, 8, 1309; (f) Adenosine agonists:
Sajjadi, F. G.; Takabayashi, K.; Foster, A. C.; Domingo,
R. C.; Firestein, G. S. J. Immunol. 1996, 156, 3435; (g)
Parmely, M. J.; Zhou, W. W.; Edwards, C. K.; Borcherd-
ing, D. R.; Silverstein, R.; Morrison, D. C. J. Immunol.
1993, 151, 389.
90 min later and the serum TNF-a activity was measured
using ELISA kits. Percent inhibition of TNF-a production
at 100 mg/kg was determined by comparison of yield with
a control to which no test compound was added. See also:
Wadsworth, S. A.; Cavender, D. E.; Beers, S. A.; Lalan,
P.; Schafer, P. H.; Malloy, E. A.; Wu, W.; Fahmy, B.;
Olini, G. C.; Davis, J. E.; Pellegrino-Gensey, J. L.;
Wachter, M. P.; Siekierka, J. J. J. Pharmacol. Exp. Ther.
1999, 291, 680.
9. Cheng, J.-F.; Ishikawa, A.; Ono, Y.; Arrhenius, T.;
Nadzan, A. Bioorg. Med. Chem. Lett. 2003, 13(21),
3647–3650.
10. Usgaonkar, R. N.; Thakor, V. M.; Jadhav, G. V.; Shab,
R. C. J. Indian Chem. Soc. 1953, 30, 743.
12. (a) Corsini, E.; Lucci, L.; Binaglia, M.; Viviani, B.;
Bevilacqua, C.; Monastra, G.; Marinovich, M.; Galli,
C. L. Eur. J. Pharmacol. 2001, 418, 231–237; (b) Squad-
rito, F.; Altavilla, D.; Zingarelli, B.; Ioculano, M.;
Calapai, G.; Campo, G. M.; Miceli, A.; Prosdocimi, M.;
Caputi, A. P. Life Sci. 1993, 53, 341–355; (c) Heguy, A.;
Cai, P.; Meyn, P.; Houck, D.; Russo, S.; Michitsch, R.;
Pearce, C.; Katz, B.; Bringmann, G.; Feineis, D.; Taylor,
D. L.; Tyms, A. S. Antiviral Chem. Chemother. 1998, 9,
149–155.
11. The in vivo TNF-a inhibition assay was conducted with
male SD rats (n ¼ 3). LPS (E. coli B055/B5, 0.1 mg/kg)
challenge was performed 2 h before the subcutaneous (sc)
administration of test compounds, which was generally
dissolved in a mixture solvent of 20% EtOH and 80%
PEG400. The pyridine-containing compounds, for exam-
ple, 22 and 27, were dissolved in saline as hydrochloric
acid salt formed in situ. The rats were sacrificed and bled